X-Search: A flexible search system for information retrieval and visualisation supporting both keyword and semantic approach

X-Search fact-file

What's the Problem?

Searching for information in legacy documents.

Objectives:

• Information overload resulting in acute need to alliviate efforts required for making relevant information available in time
• Legacy data is usually unstructured and thus difficult to query
• Knowledge in textual documents is only available for Keyword-based searching
• Keyword based search suffers from lack of precise query representation low recall compared to ontology based search
• Ontology covering all concepts and viewpoints is very expensive to build, mainten and constant modify

Towards a Solution

With X-Search, users can combine within the same query: ontology based search; keyword-based search and keyword-in-context search.

X-Search can be configured to use multiple indexes and triple stores distributed logically/geographically on multiple servers. Separation of user interface from model simplifies customisation for new applications. Keyword search is supported using using Nutch and triple store querying using Sesame with both SPARQL and SeRQL. An easy to use query interface and result visualization with highlighting and browsing features provides the user with intuitive but powerful retrieval features. Automatic graph generation (SVG. JPG. PNG) allows using retrieved information for making reports, presentations etc.